Hacker News Reader

Wanted to see where your brilliant phrase "Riscy Silver Bullet" was from so had a google, I think you are trying to start a trend ;-)

> ksec 67 days ago

> The unwritten rule of HN:

> You do not criticise The Rusted Holy Grail and the Riscy Silver Bullet.

https://news.ycombinator.com/item?id=29422032

Those RISC-V supporters are so annoying. They never seem to understand the difference between an open-source instruction set, versus an open-source chip which anyone can synthesize, versus closed-source drivers for an open-source instruction set which contains closed-source additional instructions.

RISC-V literally has no "openness" advantage over ARM other than that there's no licensing fee. Apple could make a RISC-V chip and use it in their iPhones and it would be just as locked down as their ARM chips.

One is required for the other, though - you're not going to get any good open hardware for a closed ISA, and RISC-V and OpenPOWER seem to be the two shots we've got. We practically need an open ISA to succeed as a prerequisite for open hardware to do the same.

If by open hardware you mean open source chip then yes but I’m not sure that 1. that’s a big advantage for users or 2. going to happen at the leading edge of CPU design where most of us actually do our computing. It takes real money to design and test a high performance CPU and (even for RISC-V) an open source ISA does not mean an open source chip.

Yeah as a rust user and believer even I cringe at some rust crusaders, I've seen the same for RISC V as well :(

Disclosures: I've closely followed Nvidia for 15+ years, and own stock.

I feel Jensen and Nvidia still, very strongly, wanted to do the deal. The deal increased in value to $87bn only because NVDA's stock price increased that much. The real costs to Nvidia did not increase.

ARM is a legitimate competitor to x86, which Nvidia wanted to control.

>The deal increased in value to $87bn only because NVDA’s stock price increased that much.

I get what you’re saying, but in practice, NVDA was essentially issuing x shares to buy ARM. They could do the same thing, sell them, and keep the cash. The rising stock price very much did raise the “real” cost to NVDA, it just wasn’t a simple cash charge.

Put alternatively, NVDA basically said to ARM “you’re worth x% of our company.” While that may be true at one date, a deal struck at the new market cap would be unlikely to come to the same conclusion.

Yeah, NVDA could sell the shares on the market, and buy with cash. In practice, selling that many shares tends to cause downward pressure on the stock price. But it is an option.

That said, Nvidia was still trying hard to make the deal work, even at that price. Jensen strongly believed the deal would be worth many multiples of the acquisition cost.

Yeah, without going too much into ad hominem, Jensen has been rewarded for not giving AF about what the markets think and focusing on R&D-heavy innovation (seriously listen him talk about how short-sighted the market is being about GPUs in 2012-2014).

The counterpoint ofc being (with Facebook/Meta being a great example), just because a CEO was once a visionary doesn't mean they are infallible.

The real cost to Nvidia is $87bn. Value is denominated in dollars, not nvidia stock.

If the value is denominated in dollars, why did the acquisition price go up from 40B to $87B as the price of the stock rose?

From the source:

https://nvidianews.nvidia.com/news/nvidia-to-acquire-arm-for... "The number of NVIDIA shares to be issued at closing is 44.3 million" [this is the pre-split amount]

If I agree to sell you my house in 10 years, and ten years later the house is worth 10x as much, but every other house is only worth twice as much did the value of the deal change?

You analysis is right on, and I agree with it. I predict however, the IPO will be an enormous success. Predict a valuation in the $40bn to $45bn with tops of $50bn....lets revisit this thread then ;-)

Those facts will:

- Disprove once more that story of the market efficiency/price discovery bullshit,

will also confirm that:

- Benjamin Graham, Value Investing and Warren Buffet strategies were made for other times. The times before we allowed that Central Banks hijack our income streams and personal wealth.

Disclosure: As I write this, I also do not currently hold any stocks, Mutual Funds, Indexes and other financial instruments. Did in the past.

I think you're right about this IPO, but we could be heading into a very different paradigm with the rate hikes coming.

The longer I’m in the market the more I realize George Soros was completely correct with his theories on reflexivity.

> Reflexivity theory states that investors don't base their decisions on reality, but rather on their perceptions of reality instead.

Meme stocks, Covid crash, Tesla stock, the SP500 for the last decade, NFTs. None of it is priced correctly because humans are at heart greedy, fearful, irrational. I think the trick is to figure out how to use that to make money.

There's quite a bit of academic literature on the long term overvaluation and underperformance of "growth" and especially "small cap growth" stocks, sometimes explained by their popularity as pseudo lottery tickets. Over almost all rolling 25 year periods across the past century, "value" stocks (the X% of the market highest in current earnings, cash flow, etc.) outperform the broader market including these growth stocks. Search "the black hole of investing."

I am a little bit more optimistic now knowing they are going to US for its IPO instead of LSE. Which tends to host crazy valuation with much more volume and investor willing to bet on technologies. Still, not entirely convinced it could do $40bn+. I actually think there are enormous opportunity for GPU, but they dont seems to be interested.

I also sort of disagree with Value investing as a bygone era. It is just I have yet to see any one who does value investing actually understand anything about Tech. Because it is not something simple enough like Coca Cola or McDonald, so they are much harder to evaluate.

But Yes, it will be fun to revisit this thread then. :)

And I forgot to mention about the Server part. Although it is not clear how much of an impact compared to the volume of Mobile Phone.

If you don't mind me asking, where do you keep your wealth then? Property? Commodities?

By buying ARM the owner gets control over an architecture that many describe as "the future of computing". That's much more valuable than whatever money ARM makes.

And if that control becomes too tight, RISC-V takes it place.

Let’s be real. ARM got a chance because it excelled in a very specific use case: low power. Unless RISC-V differentiates itself in a significant way, it won’t replace ARM.

RISC-V seems to me to have two major opportunities.

- Where a free ISA allows innovation and experimentation in a way that Arm would not permit.

- Where companies can shave cents off mass market products due to lower licensing fees.

Will RISC-V replace Arm in your smartphone though - seems like a big ask at the moment.

Google maintains two parallel Android platforms right now, so the idea of porting everything over to a third is not _too_ far off. There's a lot of effort in getting JIT up-to-speed on RISC, but most everything else is a recompile away.

And JIT can come in later. But definition it's making runtime decisions on code generation, so being supported by optional extensions isn't the end of the world.

I’m old enough to remember Intel throwing billions at getting x86 into smartphones.

Just because there is a port available doesn’t mean the hardware and ecosystem will follow.

Whether it'll make it into a phone is a whole other matter, but there has been at least one RISC-V port of AOSP:

https://www.xda-developers.com/android-risc-v-port/

> RISC-V seems to me to have two major opportunities.

> - Where a free ISA allows innovation and experimentation in a way that Arm would not permit.

I would say that the two major innovations of RISC-V are:

- Composability.

- Extensibility (that also fosters experimentation).

Each ARM version X.Y comes with a license and with a condition: «thou shalt taketh either everything or nothing»; RISC-V has neither a license nor strings attached and is «thou may useth only what thou needeth». Anyone who does not need SIMD or the vector processing can do away with them in the silicon. It saves the die size[0] and costs. This is the composability aspect.

Extensions planned out and/or reserved for future use (e.g. 128-bit extension of RISC-V), and the ability to add custom instructions to experiment with (and possible contribute to the RISC-V forum). This is the extensibility aspect.

Other existing open ISA (MIPS, POWER SuperH J-2? – unsure on the status of the last one) do not offer composability and extensibility capabilities, which leads me to reason that it is the major innovation(s) of RISC-V.

[0] Albeit it is somewhat fraught with the danger of creating a highly fragmented RISC-V CPU market where the software has to accommodate every possible permutation of RISC-V extensions and inject an emulation layer for extensions missing in the die.

ARM also addressed code density with Thumb, which required licensing Super-H IP.

Then AArch64 was designed very late, but seems sufficient to have (currently) taken the prize for the top supercomputing architecture.

There is architectural flexibility here, in addition to low power.

You seem to be knowledgeable about this, so I wanted to ask - is ARM more efficient (perf/watt) than Risk-V? (I know there are lots of factors like frequencies/voltages etc but keeping everything else constant)

That is really a loaded question.

ARM bigLITTLE exists to address different efficiencies that are exploited in different runtime states. Is big better than LITTLE? Sometimes yes, sometimes no.

If there is no definitive answer within the heterogenous cores of a single ARM implementation, then how can it be compared to RISC-V?

Ok, I'll try again.

For the same sized single-core chip (x mm squared), running at the same voltage and same clock speed, on the same manufacturing process node, is there a (significant) difference between ARM and Risk-V?

(I'm assuming that core size, voltages and frequencies are the biggest differences between the big and LITTLE cores. And yes of course this is a funky theoretical question so a theoretical answer is perfectly fine :)

The answer (for now) appears to be no. Also consider that ARM will not stand still.

"According to SiFive, it is working on the design of a new processor that will be 50% more powerful, and that it will be able to surpass the Cortex-A78 of ARM."

https://techunwrapped.com/risc-v-starts-to-put-pressure-16-c...

https://news.ycombinator.com/item?id=29418153

> For the same sized single-core chip (x mm squared), running at the same voltage and same clock speed, on the same manufacturing process node, is there a (significant) difference between ARM and Risk-V?

Very many. And they are at the CPU implementation level.

The instruction decoding buffer depth, the register file size, the implementation of the out of order execution, the TLB depth, the data bus width, separate hardware assist engines etc are amongst many other conscious and decisive design decisions that can differentiate separate designs. Even at the same (citing verbatim) «voltage and same clock speed, on the same manufacturing process node» the performance can vary wildy.

The M1 CPU family is a very wide implementation in nearly every aspect, with a very large L1 cache (both, I and D, caches), reasonably large L2 cache, a (reportedly) 380 register wide register file, 512-bit wide data bus (in the M1 Max) plus extras, hence why it outperforms competing ARM64 designs running at similar voltages and frequencies. The same is true for POWER CPU's (i.e. compare POWER9 vs POWER10 designs) – POWER10 is 2.6x faster than POWER9 despite running at similar frequencies – due to a substantial design overhaul.

Holistic system design matters, not the ISA alone.

>difference between ARM and Risk-V?

Which is the problem often with these sort of discussion. ARM of what ( You didn't state its version ) ? RISC-V of what? The ISA in itself? Or the actual implementation in itself. If it is implementation, which implementation? And by difference, what sort of difference, what kind of performance are you specifically looking at? Or you just want a GB5 result? ( One could argue there are compiler differences in GB5 results )

The only undisputed truth would be RISC-V at its bare minimum has smaller instruction than any ARM variation currently on the market. But then its competitor should have been something like Arc, and not ARM.

The ISA doesn't have that much influence on efficiency (other than, like, x86's insane-length encoding requiring spending more power on decoding).

"About like ARM but without licensing fees and not controlled by ARM (the company)" could easily be a sufficient differentiator, especially if ARM the company starts acting out.

> I much rather bet on Intel if I had the money.

The FTC objecting to Nvidia and ARM's deal on an antitrust basis [0] makes Arm toxic to Intel and AMD.

[0] https://www.ftc.gov/news-events/press-releases/2021/12/ftc-s...

Intel have been ARM licensees and made ARM chips before (StrongARM which they acquired, XScale which they developed).

They've always managed to display ambivalence about it. We should be glad they never owned ARM and hopefully never will; they'd have killed it with indifference.

But Intel has a P/E of 9.91, they must be teetering on the brink of bankruptcy!

It's really funny how the market thinks that Intel isn't in the game anymore.

They may not be as energy efficient as M1 or ARM chips but they certainly show that they are still a competition and invest a lot of money to improve their performance.

The thing is, the market can be right without knowing why it is right.

The market thinks that in some fixed, relatively near point in the future we will look back at now and say "Intel was already hosed, it just didn't know it".

It may be that the market is currently only expressing interest in the hopes of others that this particular future will come to pass, but that is the only way it will.

The market used to think AMD was hosed. About 5 years ago, the stock was worth less than 2$. Now its more than a 100$. ¯\_(ツ)_/¯

Right but the key difference is that post-recovery AMD is fabless, because they spun off their fab business.

AMD's stock can jump like this, because its value is based more closely on its designs.

Intel is not taking that approach. One amazing chip design isn't going to massively change their fortunes, and they are heavily sunk in x86, which is not the chip series the market is most interested in.

One meaningful path forward for Intel is to split in two, precisely so that the x86 design business can be liberated by being fabless, and the fab businesses can get on with making unsexy larger-process chips for everyone in America who needs them.

But Intel seem to be resisting this.

When AMD was at 2$, the story was that their chips designs are shit, slow, overheating etc. And since they had just flogged off their foundry they didn´t have access to any useful process, thus the chips were more slow and hot and power hungry. They also tons of debt, and basically no assets, and no way to compete - while Intel had the best chip designs and the best foundries.

We'll see what the story will be going forward.

But they really haven't. We've not seen hide nor hair of the new process, only iterative improvements on the current process, and the attempt at big.LITTLE x86-style doesn't seem to have worked out terribly well.

ARM has taken out the low-power market that Atom and Pentium classes were aiming at. AMD APUs have long been a favorite of mid-power devices like gaming consoles and are now a desktop enthusiast favorite as well. The i5/i7 class is outmatched in every metric ($/FLOP, W/$, and W/FLOP) by the Ryzen series. IBM and Azure placed their big chip refresh orders with AMD as well, so the hyperscaler market is in danger. The only place Xeons are still superior is the scientific and high computing space, and Threadripper is making inroads there as well.

Intel still has great sales from existing cashflows and OEM agreements, but if you were buying a computer in 2022, why choose Intel? Why would its customers? I think they're going to have to come up with something new and soon.

> The i5/i7 class is outmatched in every metric ($/FLOP, W/$, and W/FLOP) by the Ryzen series.

This isn't really true though. Alder lake performance is better than Ryzen 5x at pretty much all price points. Alder lake's electricity usage is rough, but desktop users don't care. Now ryzen is a year behind intel, and who knows how their next offering will fare, but saying they've outmatched intel in every metric just isn't true.

> and are now a desktop enthusiast favorite as well. The i5/i7 class is outmatched in every metric ($/FLOP, W/$, and W/FLOP) by the Ryzen series.

maybe a couple years ago, at this point Intel is dominating the gaming market, competing in the productivity market despite having fewer cores (i.e. more potent cores), are at least matching Zen3 efficiency (except i9, but you said i5/i7) if not beating (particularly at lower power budgets due to infinity fabric overhead). Ryzen has absolutely lost favor with the enthusiast market due to the significant price increases (in some segments, a 50% increase over last generation) and the refusal to chop prices to remain competitive with intel (justified with "but they can make more money selling Epyc", which is great for them but it doesn't make their products competitive for me as a consumer either).

And AMD has completely abandoned the value market... i3 12100F is listed for $110 at Best Buy (iGPU version is $140), and Intel is in the process of rolling out Celerons for even cheaper rigs (will be nice for HTPC with that iGPU). AMD's cheapest processor has a $300 MSRP and they won't budge on pricing at all given their focus on the server market.

go look at a tech enthusiast forum and nobody is recommending Ryzen anymore. It's overpriced, it no longer has a performance or efficiency advantage (in fact it's rather flopped the other way), and AMD would rather address the server market instead.

this has really shifted in the last 6 months since Alder Lake launched, but really there was a significant niche for recommending 10-series as a budget option given the significant price advantage over AMD. The thing that changed with Alder Lake is that Intel moved back into the performance lead (and matched efficiency) and prices remained the same - leaving them both better and cheaper than AMD. Thing have changed, people don't recommend AMD anymore unless you have a very specific need (like you're buying a 5950X).

Also APUs in particular are rather a flop on desktop. AMD didn't even bother releasing a 7nm APU to the consumer market until... September last year? Up until then it was all 12nm quad-core garbage - and that's GloFo 12nm, so more like Intel 22nm. On top of that, the APUs also had reduced IPC compared to the CPUs of their equivalent architecture (due to less cache). All of the hype around Ryzen was around the non-APU enthusiast chips with no onboard graphics, the APUs were never popular in the desktop segment.

The APUs are nice if you are building a SFF rig with no discrete GPU... but they have a lot of compromises - reduced IPC, PCIe 3.0, and up until September they didn't have any that weren't limited to PCIe 3.0x8 (5000G series moves this to 3.0x16). They have no HDMI 2.1 support, no HDMI VRR support, no AV1 support, and the worst hardware encoder (i.e. Teams videoconferencing quality/etc) on the market. If you really want a gaming-focused SFF rig with iGPU only, they are the best available, but they were never popular, and AMD never really pushed them the way they did the enthusiast chips.

(HEDT is another segment that's been abandoned... AMD still has not released Zen3 processors, it's been 16 months now, and who knows whether AMD will try the same "release it for pro but not for consumer" crap again like they did with the APUs.)

Because Intel is / will no longer be a x86 / Chip company only anymore. They are pivoting to Foundry. Just like Andrew Grove pivoted Intel from DRAM to microprocessor.

I expect them to be at least $400bn to $600bn market cap by the end of 2030. Which is 2x or 3x of today's price.

Not exactly super high growth, but if you are a money manager with too much cash ( which is often the case today ) it is not a bad investment.

Have their profitability prospects taken a hit because of whatever happened at ARM China?

We dont know because Softbank no longer breaks out those number in 2018/2019. But they do give some insight into the level of growth so we could extrapolate from previous year or quarters.

And generally speaking Softbank ( and ARM ) refuse to comment on anything about ARM China.

But that is a good point, it is a potential risk or at least need to be mindful in its valuation.

FWIW Arm still files accounts in the UK (apparently for the Group).

https://find-and-update.company-information.service.gov.uk/c...

Latest is 2020/21. Profits down to £7m due to higher R&D and other expenses.

ARM licenses are mostly for complete cores and associated IP blocks that work on your process. No reason they can't make RISC-V ones.

"perceptual licensing agreement" ? Could you clarify in what sense the agreement is "perceptual"?

pretty certain he means perpetual

If Activision/Blizzard is worth $65bn to Microsoft for it's IP then $84bn is still a steal for ARM?

But that was a good investment for its long term Gamepass/Xbox plan. Or are you saying Microsoft can buy it?

Provided the ARM juggernaut is expected to streamroll Intel's business in the coming decade, this acquisition does make sense for Intel... this could be their Facebook+WhatsApp moment.

If Nvidia was blocked from acquiring ARM, then I certainly think that Intel would be blocked as well. Plus, Intel does not need to actually acquire ARM to start making ARM processors. For example, Apple has done just fine licensing ARM IP just like everybody else.

Intel indeed already is a large ARM licensee (and for a while even was the largest one, funnily enough)

To be clear, Apple is not licensing ARM IP "just like everybody else". They have a very different relationship, being a founding shareholder, and then acquiring an "Architecture License" from ARM after also acquiring PA Semi according to Wikipedia.

So Apple Silicon is not the same as someone else licensing ARM IP.

Source? According to Wikipedia, quite a few companies have an architectural license with ARM, including Intel and AMD. Maybe Apple gets a discount?

Apple may have been a founding member, but they don’t own ARM, and I’m assuming that ARM still wants to make a profit off of its licenses to Apple.

If Nvidia gets torpedoed by regulators, it’s hard to see how Intel stands a chance

Intel has significant experience in managing cross-license agreements (see x86, x64).

This is getting into fantasy chip designer territory, but what are the chances that Intel makes a hybrid x86-ARM chip?

As I understand it the value for Nvidia came from adding Nvidia GPU IP into the products that AMD sells to customers.

Nvidia would have expanded the market from their own products to licenses.

Your disclosure intrigues me. What do you own?

NFTs only